Her finner du en samlet, strukturert oversikt over hvordan GoRabbit fungerer – for kunder, tekniske brukere og governance/styring.
Controls all publishing jobs.
Publisher is the core orchestrator of all publishing jobs in GoRabbit. RabbitEngine generates queues, and Publisher receives these queues and determines when, where, and how each job will run. It coordinates timing, priorities, proprietary internal mechanism handled automatically by the system.
Connects proprietary internal mechanism handled automatically by the system.
Publisher is the midpoint between RabbitEngine and RabbitAgent. It reads proprietary internal mechanism handled automatically by the system. Without Publisher, the system would lack coordination, timing, and flow.
Reads proprietary internal mechanism handled automatically by the system.
Publisher reads: (1) publish_queue.json, (2) warmup_queue.json, (3) job_config.json, (4) campaign metadata, (5) group information from proprietary internal mechanism handled automatically by the system. These are used to organize all active and upcoming publishing jobs.
Status, logs, and dispatch files to proprietary internal mechanism handled automatically by the system.
Publisher generates: (1) job_state.json, (2) Slack logging, (3) updated proprietary internal mechanism handled automatically by the system. Everything is sent back to proprietary internal mechanism handled automatically by the system.
Handles all timing and order.
Publisher reads scheduled timestamps from the proprietary internal mechanism handled automatically by the system. It respects proprietary internal mechanism handled automatically by the system. Publisher ensures publishing happens precise and safely.
Balances multiple campaigns simultaneously.
Publisher contains a priority system that ensures campaigns do not interfere with each other. Each campaign has its own section of the proprietary internal mechanism handled automatically by the system. Important or premium campaigns may get higher priority.
Dispatches jobs via proprietary internal mechanism handled automatically by the system.
When the execution time arrives, Publisher creates a dispatch proprietary internal mechanism handled automatically by the system. The proprietary internal mechanism handled automatically by the system. This proprietary internal mechanism handled automatically by the system.
Yes — publishes logs to Slack.
Yes. Publisher sends Slack notifications for: (1) job starts, (2) completed jobs, (3) errors, (4) risks, (5) restrictions, (6) proprietary internal mechanism handled automatically by the system. Slack acts as the cockpit for the entire publishing system.
Stops publishing upon risk.
Publisher pauses publishing if: (1) proprietary internal mechanism handled automatically by the system. Publisher places jobs in 'hold' mode and alerts Slack immediately.
Parallel capacity = multiple Agents.
Publisher can manage many jobs in parallel, but each proprietary internal mechanism handled automatically by the system. To increase capacity, multiple RabbitAgent instances are started. Publisher distributes jobs across agents and prevents overload through an internal load-balancing mechanism.
Combines FIFO and weighted priority.
Each job receives a priority from the proprietary internal mechanism handled automatically by the system. Publisher uses a combination of FIFO and weighted priority to ensure optimal flow.
Time zones and timeframes per campaign.
Publishing windows are timeframes defined per campaign, such as 'weekdays 08:00–22:00' or 'weekends 10:00–20:00'. Publisher ensures jobs never run outside these windows. This is essential for customers wanting controlled exposure.
Errors are mild, moderate or critical.
Errors are handled in three levels: (1) Mild — the job is retried later. (2) Moderate — the job is placed on hold. (3) Critical — the entire campaign is paused. Publisher uses signals from proprietary internal mechanism handled automatically by the system.
Designed for stable, high performance.
Publisher ensures performance through: (1) proprietary internal mechanism handled automatically by the system.
proprietary internal mechanism handled automatically by the system.
proprietary internal mechanism handled automatically by the system. Publisher receives them, executes them, and reports back status, errors, and results. proprietary internal mechanism handled automatically by the system. Publisher = execution. Together they form the complete publishing pipeline.
Coordinates all publishing.
Publisher is GoRabbit’s publishing command center. It ensures proper timing, control, risk analysis, and coordination between proprietary internal mechanism handled automatically by the system.
Queues, dispatch, logs, and state files.
Publisher uses a simple and robust proprietary internal mechanism handled automatically by the system.json), and /config/ (timing rules and limits). This ensures transparency and easy debugging of the publishing flow.
Recovers without losing jobs.
Publisher always reads job_state.json on startup. If the system crashed, it checks which jobs were ‘running’, ‘pending’, or ‘completed’. Running jobs are marked as ‘retry required’, and the system resumes seamlessly without data loss.
Dispatch proprietary internal mechanism handled automatically by the system.
Publisher generates a dispatch proprietary internal mechanism handled automatically by the system. proprietary internal mechanism handled automatically by the system. The handover is fully proprietary internal mechanism handled automatically by the system.
Yes — scales to large volumes.
Yes. Publisher is built to handle thousands of daily posts. The system scales by running multiple agents, and Publisher automatically balances job loads. The proprietary internal mechanism handled automatically by the system.
FIFO + priority scoring.
Publisher uses a combination of FIFO (first-in-first-out) and priority scoring. Jobs with identical priority follow proprietary internal mechanism handled automatically by the system. proprietary internal mechanism handled automatically by the system.
Yes — strictly enforces rate limits.
Yes. Publisher includes natural pauses, 150–210 proprietary internal mechanism handled automatically by the system. If the proprietary internal mechanism handled automatically by the system.
Yes — fully customizable per campaign.
Yes. Each campaign may define: (1) max groups per day, (2) custom time windows, (3) no-night posting rules, (4) regional filters, (5) category targeting. Publisher enforces all of these automatically.
Controls the entire publishing regime.
The Publisher configuration contains: (1) parallel proprietary internal mechanism handled automatically by the system. The config controls the entire publishing flow.
Yes — continuous health checking.
Yes. Publisher checks: (1) proprietary internal mechanism handled automatically by the system. Publishing stops if any critical failure is detected.
Atomic writes protect queues.
Publisher never locks files but uses atomic writes and temporary .tmp files. This ensures that proprietary internal mechanism handled automatically by the system. proprietary internal mechanism handled automatically by the system.
Default = server timezone.
Publisher uses the server timezone by default. Customers may define custom timezones per campaign, and Publisher automatically translates all scheduled timestamps to server-proprietary internal mechanism handled automatically by the system.
Yes — fully transparent in files.
Yes. Publisher is built with maximum transparency. Everything is written to files: queues, state, dispatch files, logs. Each job has a unique ID and full traceability. The entire system can be debugged without a database or GUI, just by inspecting files.
Pause mode on risk.
Hold mode is activated when Publisher detects risk or errors. Jobs are paused without being deleted. This prevents unsafe publishing. Hold mode is manually released via Slack or the GoRabbit console.
Stops night-time publishing.
Night mode prevents publishing during customer-defined nighttime windows (e.g., 23:00–06:00). Many customers prefer this to reduce risk and ensure posting aligns with audience activity patterns.
Not yet — coming in v3.
Not in the initial release. Webhook support is planned for GoRabbit v3 for real-time API integration. Currently, all communication is proprietary internal mechanism handled automatically by the system.
GoRabbit’s publishing controller.
Publisher is the intelligent coordinator ensuring everything the proprietary internal mechanism handled automatically by the system.
Automatically selects the least-loaded proprietary internal mechanism handled automatically by the system.
Publisher evaluates all available RabbitAgent instances and selects the one with the lowest load, fastest response, and available capacity. The system avoids overload using an internal load-balancing mechanism based on CPU, memory, uptime, and active jobs.
Yes — full validation before dispatch.
Yes. Publisher validates that: (1) text exists, (2) image exists, (3) group ID is valid, (4) tokens and cookies are available, (5) the job is not a duplicate, and (6) the publishing timestamp is correct. If anything is missing, the job is halted before being dispatched.
One retry — then hold.
Publisher retries a failed job once. If it fails again, the job is moved to ‘hold’ and the proprietary internal mechanism handled automatically by the system. This prevents endless loops and reduces the risk of triggering Facebook restrictions.
One job per account at a time.
Yes. Publisher enforces concurrency limits per account to prevent patterns that Facebook may detect as automation. By default, only one job per account may run at any given time, regardless of the number of agents.
Yes — supports job chaining.
Yes. Some jobs rely on earlier jobs, such as proprietary internal mechanism handled automatically by the system. Publisher marks these as ‘dependent jobs’ and ensures they run in exact order regardless of proprietary internal mechanism handled automatically by the system.
State stored in job_state.json.
Publisher stores its state in job_state.json and a dedicated state directory. This includes job IDs, status, timestamps, priority, retry state, and any errors. On restart, everything is reloaded without data loss.
Halts on proprietary internal mechanism handled automatically by the system.
If the proprietary internal mechanism handled automatically by the system. Slack receives a critical alert. No jobs are deleted, and the system will not publish again until an operator validates disk integrity.
Fails safely and halts further posting.
If the proprietary internal mechanism handled automatically by the system. Publisher marks the job as ‘retry required’. If the network is fully down, all jobs enter hold mode. The system avoids publishing without network to prevent partial failures.
Coordinated flow between proprietary internal mechanism handled automatically by the system.
Full Publisher flow: (1) proprietary internal mechanism handled automatically by the system.
Very low resource usage.
No. Publisher is extremely lightweight. It primarily works with small JSON files and runs short async checks every proprietary internal mechanism handled automatically by the system. Even on low-spec servers, Publisher typically uses under 2% CPU.
Yes — several fixed system limits.
Yes. Publisher has hard limits on: (1) max jobs per proprietary internal mechanism handled automatically by the system. These prevent overload and are part of core system safety.
Yes — extensive failsafe system.
Yes. Publisher can globally halt all jobs on critical errors, lock specific campaigns, freeze proprietary internal mechanism handled automatically by the system. This protects customer Facebook pages and posting history.
Yes — customer sees simplified view.
Yes, but in simplified form. Customers see: (1) progress, (2) upcoming posts, (3) job status, (4) any stops. The technical details (queues, state files, failsafe logic) are hidden to keep the UI clean.
Yes — AI planned for v3.
Yes. GoRabbit v3 will include AI-driven prioritization, predictive risk scoring, and smart publishing time selection based on engagement history. This will further optimize posting.
Several major extensions planned.
Planned features include: (1) webhook support, (2) API-based publishing, (3) intelligent campaign scheduling, (4) visual proprietary internal mechanism handled automatically by the system. These are part of the GoRabbit v3–v4 roadmap.
Executes everything proprietary internal mechanism handled automatically by the system.
Publisher is the system that ensures everything proprietary internal mechanism handled automatically by the system.
The module that publishes posts.
The Publisher module is the part of GoRabbit that actually publishes the content into Facebook groups. It receives ready proprietary internal mechanism handled automatically by the system. Publisher is what produces the visible real-world results.
proprietary internal mechanism handled automatically by the system.
The proprietary internal mechanism handled automatically by the system.json. The Publisher carries out the actual work. In short: proprietary internal mechanism handled automatically by the system. Publisher = action and execution.
Publisher’s to-do list.
campaign_queue.json is the proprietary internal mechanism handled automatically by the system. Each entry contains group_id, group_name, image_path, teaser, proprietary internal mechanism handled automatically by the system. This proprietary internal mechanism handled automatically by the system.
Supports text and image posts.
Publisher can publish: (1) image posts, (2) text posts, (3) teaser posts, (4) combined image + teaser, (5) fallback text if images are not allowed. Future versions will support carousel posts, link posts and reels.
Realistic and varied timing.
Publisher uses proprietary internal mechanism handled automatically by the system. This makes Facebook perceive activity as normal human behavior.
Automatically stops on risk.
Publisher analyzes rejections, error messages, popups, ‘slow down’ notifications, CAPTCHA-like patterns and unusual loading behavior. If risk is detected, publishing stops temporarily and the proprietary internal mechanism handled automatically by the system.
Retry up to three times.
The Publisher marks the post as failed and schedules a retry with a randomized proprietary internal mechanism handled automatically by the system. A maximum of three attempts is made. If all fail, the group is excluded from the campaign and risk is reported to Slack.
Full live reporting.
The Publisher reports each step of the process: when posting starts, completes, fails, triggers risk or gets excluded. This gives high transparency and makes it easy to follow the campaign in real time.
Fallback strategies for edge cases.
Edge cases include popup blockers, unknown modal windows, group changes, temporary network errors and session timeouts. The Publisher has fallback strategies for all scenarios and always tries mild retry before a full stop.
Automatic re-login.
Publisher uses stored session cookies. If a cookie expires or becomes invalid, the proprietary internal mechanism handled automatically by the system. Changes are updated in tokens.json and reported via Slack.
Yes — multi-proprietary internal mechanism handled automatically by the system.
Yes. Publisher supports multiple parallel proprietary internal mechanism handled automatically by the system. This allows massive scalability for high-volume clients.
Skips on risk or rule changes.
The Publisher skips groups that: (1) require admin approval, (2) no longer allow posting, (3) trigger risk signals, (4) do not support image posts, (5) return unusual errors. All skips are logged to Slack.
Campaign finalized and archived.
When all tasks in the proprietary internal mechanism handled automatically by the system.json, sends a completion message to Slack, and marks the campaign as completed in the proprietary internal mechanism handled automatically by the system.
Publisher = actual publishing.
The Publisher is the proprietary internal mechanism handled automatically by the system. It uses proprietary internal mechanism handled automatically by the system. Without Publisher, no posting happens.
Full control via Playwright.
Publisher runs inside controlled Playwright environments and has full control over browser actions: opening tabs, scrolling, input, uploads, clicking, handling modal windows and dynamic components. All actions use human-mimicked timing so Facebook does not flag the behavior.
To wait for Facebook’s load times.
Facebook rendering is non-deterministic. Elements appear in unpredictable order and timing varies. Publisher uses adaptive render delays that detect when input fields, buttons and upload containers are actually ready before performing actions. This massively increases stability.
Uses robust and relative selectors.
Facebook continuously changes IDs, class names and DOM structure. Publisher therefore uses a robust strategy based on relative positions, visibility, accessibility, text anchors and fallback selectors. This makes the system resilient against UI changes.
Cleans and prepares text for posting.
Facebook has built-in filters that may strip formatting or block non-standard characters. Publisher cleans the teaser text, replaces problematic unicode characters and formats line breaks correctly before posting.
Group excluded on admin block.
If a group requires admin approval or posting is disabled, Publisher immediately aborts the group and marks it as admin_blocked. The group is excluded from the rest of the campaign. Slack receives a detailed notification.
Popup detection + automatic handling.
Popups such as ‘Are you sure?’, ‘Review post’, ‘Try again later’, and ‘Something went wrong’ are detected via pattern matching in the DOM. Publisher analyzes the popup, chooses the correct action (close, retry, skip) and reports to Slack.
Scrolling = natural behavior + validation.
Scrolling is used to mimic human behavior: (1) scroll to activate the group feed, (2) scroll before posting to find the input field, (3) scroll after posting to verify that the post is visible. This reduces errors significantly.
Image → text when needed.
If an image post fails or is not allowed in a group, the Publisher falls back to plain text posting. This ensures the campaign still reaches the group with maximum compatibility and minimal risk.
Retry + pause on network failure.
The Publisher evaluates network condition at each action. If the network drops or latency becomes too high, a retry sequence is triggered. If the issue persists, the campaign is paused and Slack is notified.
DOM monitoring for load stalls.
Publisher monitors DOM changes. If Facebook stops loading new content within a defined threshold, Publisher triggers a reload or fallback. This protects against infinite loading loops.
Yes — optional screenshot log.
Yes. The Publisher can take screenshots after the post is confirmed published. This is useful for debugging, documentation and customer proof. Screenshots are stored temporarily and deleted after a defined retention period.
Typically 150–240 sec per post.
Publishing speed varies based on proprietary internal mechanism handled automatically by the system. A typical range is 150–240 seconds between posts. The system can go faster, but this increases risk and is used only in tests.
Campaign stops and cooldown starts.
If Facebook blocks or limits the account, Publisher immediately stops the campaign. The proprietary internal mechanism handled automatically by the system. A new proprietary internal mechanism handled automatically by the system.
proprietary internal mechanism handled automatically by the system.
When multiple agents run simultaneously, the proprietary internal mechanism handled automatically by the system. Each proprietary internal mechanism handled automatically by the system. The system scales linearly with the number of agents.
DOM check to confirm posting.
Publisher scans the DOM after posting to detect confirmation elements such as ‘Your post is now live’, ‘Posted just now’, or the appearance of new nodes in the feed. If validation fails, a soft retry is attempted. Repeated failures trigger Slack alerts.
Short pause after posting.
Post cooldown is a safety delay inserted after a post is published. It ranges from 5 to 20 seconds and simulates human behavior while giving Facebook time to process the content.
Yes — and campaign halts immediately.
Yes. The Publisher detects patterns that indicate CAPTCHA, verification prompts, identity checks or other friction. When detected, the campaign is halted immediately and a high-risk alert is sent to Slack.
Automatic proprietary internal mechanism handled automatically by the system.
Publisher monitors session cookies for expiration or invalidation. When a refresh is required, RabbitAgent performs a new login. The updated cookies are saved to tokens.json and logged to Slack for traceability.
5 different publishing modes.
Publisher supports: (1) Strict Mode (safe/conservative), (2) Balanced Mode (default), (3) Aggressive Mode (testing only), (4) Multi-proprietary internal mechanism handled automatically by the system. Mode is selected by the proprietary internal mechanism handled automatically by the system.
Gradually increasing retry intervals.
On the first failure a short retry is used. On the proprietary internal mechanism handled automatically by the system. On the third failure the group enters a 30–90 minute backoff. On the fourth failure the group is excluded from the campaign. This reduces risk exposure.
Problematic groups auto-blocked.
If the proprietary internal mechanism handled automatically by the system. The group is excluded from all future campaigns until manually reviewed.
High-priority groups always used.
Whitelisted groups are groups that should always be published to regardless of risk or fallback. This is used only for owned pages, client-specific groups and internal feeds with high trust.
150–700 MB per proprietary internal mechanism handled automatically by the system.
Memory usage varies between 150 MB and 700 MB per proprietary internal mechanism handled automatically by the system. The system is optimized to avoid memory leaks and releases resources between posts.
20–45% during active posting.
CPU consumption is moderate: 5–15% idle and 20–45% during posting. This comes from Playwright’s DOM inspection and image handling routines. The system performs well even on low-cost proprietary internal mechanism handled automatically by the system.
proprietary internal mechanism handled automatically by the system.
Communication operates through proprietary internal mechanism handled automatically by the system. Publisher initiates actions, the proprietary internal mechanism handled automatically by the system. This eliminates the need for complex APIs.
Tasks locked during execution.
Publisher locks proprietary internal mechanism handled automatically by the system. When an proprietary internal mechanism handled automatically by the system.json. This prevents two agents from posting to the same group simultaneously.
Yes — High-Precision mode.
Yes. High-Precision Mode enables additional controls: longer render delays, stricter validation, extra post verification and more DOM checks. Used only for high-priority clients or pilot programs.
High-level overview only.
Customers receive a summary in the GoRabbit.no UI: number of groups posted, excluded groups, remaining proprietary internal mechanism handled automatically by the system. They do not see technical details like selectors, DOM structure or internal logs.
Full posting history stored.
Yes. Publisher stores history in the archive/ directory with metadata for each post: timestamp, group, result, teaser, image, errors and risk. The history can be exported on demand.
Yes — global compatibility.
Yes. Publisher is fully language-agnostic and works in all markets where Facebook groups exist. The system can adapt to regional time zones, regional risk and language-specific teasers.
Publisher = the proprietary internal mechanism handled automatically by the system.
The Publisher is the part of GoRabbit that performs real Facebook group publishing with human-mimicked behavior, risk control, variation and robust error handling — one of the most critical modules in the system.